Dashboard Design for Churn Prediction Model

Client

A global diagnostic systems manufacturer specializing in diabetes care and glucose monitoring systems.

Business Objectives

Increase the active user base for digital health monitoring solutions
Improve patient measurable outcomes through better engagement
Enhance value delivery through cloud based analytics
Strengthen strategic partnerships and innovation capabilities

Proposed Solution

Design and implement a user friendly dashboard that enables teams to:

Identify customer churn patterns using predictive analytics
Understand churn drivers through intuitive data visualizations
Act on insights with targeted interventions and recommendations

UX Process

Secondary Research

Understanding What is Churn?

Churn in simple terms is, when customers stop using a product or service.

Customer churn versus revenue churn diagram

Churn Rate Formula

The ratio of customers lost during a given period to the total number of customers at the start of that period.

App Store & Play Store Reviews

To better understand real user frustrations and motivations, we conducted secondary research in collaboration with the data team. We did the following:

Collected and analyzed user reviews from App Store and Play Store
Applied AI and ML models to categorize and generate positive and negative word clouds
Identified key problem areas and user pain points

Positive User Reviews (79%)

Overall good to track insulin dosage and glucose levels.
Accurate 90 day prediction, almost near to hb1ac levels
Good look and feel of app.

Negative User Reviews (21%)

Meter and App sync issues reported by most.
Data sharing difficulty with physicians
Difficulty in editing data, lack of data precision.

Sample app store and play store reviews with sentiment tags

These insights helped us define clear UX opportunities to improve on what users actually care about.

Competitive Analysis

We examined three major competitors: OneTouch Reveal, Glucose Buddy, and mySugr.

Competitor study of OneTouch Reveal, mySugr, and Glucose Buddy

The following were the improvement areas derived from competitor study:

App customization and personalization features
Adding photo feature to remember and share
Coaching users by providing insights to reach target goal
Providing nutritional information, safe and spike levels both
Data precision by adding decimal parameter for glucose readings
Crashing and app syncing issues

User Persona

A key user we designed for is Danny Roy, a Data Scientist who analyzes churn weekly.

"Time to make data ethics more practical and actionable."

We created user persona for the users of the dashboard, data scientists, product owners, and marketing lead, to identify the key focus area responsibilities of the user.

Danny's journey helped us build a role aware dashboard, tailored for clarity and quick decision making.

KPI Mapping

After creating user personas for each user type we mapped the relevant KPIs for each user persona.

Low Fidelity Wireframe Design

Created concept dashboard and coordinated with stakeholders to get continuous feedback and iterated on low fidelity wireframes.

Low fidelity wireframe of the churn dashboard layout

High Fidelity Dashboard Design

The final dashboard was designed to bring clarity, depth, and actionability to the churn prediction process. It clearly communicates key drivers of churn surfaced directly at the top: meter errors, login issues, data sync problems, privacy concerns, and app version related drops, especially post upgrades.

Left section of the dashboard shows:

Churn prediction rate
Top 3 churn factors, updated dynamically
iOS vs Android churn risk comparison

Right section highlights:

Total users gained and lost
App usage by features
Churn rate using actual formula based calculations
Key drivers of user churn
Churn prediction by diabetes type, age group, and geography (map view)

High fidelity churn prediction dashboard

Together, these visualizations help internal teams identify trends, uncover user pain points, and take proactive actions to reduce churn.

Conclusion

This case study was a great example of how user experience design can make complex data actionable and human friendly. It combined real user feedback, AI and ML driven insights, and clear visual storytelling.

Also, aligned multiple teams, from data scientists to support leads, with a shared, easy to understand view of user behavior.

Most importantly, it helped move the conversation from "What's going wrong?" to "Here's what we can fix, and how."